AI Reading for Monday January 6

Jan 06, 2025

Sam Altman Q&A on ChatGPT’s First Two Years, Elon Musk and AI Under Trump - Bloomberg

"We are now confident we know how to build AGI as we have traditionally understood it." samaltman.com

$200/month ChatGPT Pro users are going ham. I feel like you could point advanced voice mode with video at stuff 24h a day and have it look for stuff. Wait til someone specific passes your window and drop a water balloon on them.

DeepFace is a lightweight face recognition and facial attribute analysis (age, gender, emotion and race) framework for python. It is a hybrid face recognition framework wrapping state-of-the-art models: VGG-Face, FaceNet, OpenFace, DeepFace, DeepID, ArcFace, Dlib, SFace and GhostFaceNet. - GitHub

Reports of AI running out of data have been greatly exaggerated. - Business Insider

AI training is currently very sample inefficient, like a student that has to be told the same thing many times in different ways before it sinks in. The larger the model, the worse the problem. But since AI is good at understanding and generating text, it can generate those variations itself as it trains.

To Interact With the Real World, AI Will Gain Physical Intelligence - WIRED

Arbor’s ‘reasoning’ chatbot analyzes text and extracts investment ideas to enhance fundamental stock analysis, company says. - South China Morning Post

Using AI for workouts ... tell Advanced Voice Mode with video to act as your personal drill sergeant - PCMAG

How Are Companies Using AI Agents? Here’s a Look at Five Early Users of the Bots - WSJ

Mat Honan on AI and the future of Google Search - MIT Technology Review

Mirror mirror on the wall…AI smart mirror to monitor your health, talk to your wearables and scale. - ZDNET

Robot vacuum will pick up your socks after you without busting your chops - Engadget

As the twenty-first century began, human evolution was at a turning point. - Reddit

Klarna's CEO says AI is capable of doing his job — and it makes him feel 'gloomy'. Don't we all. - Business Insider

Buterin proposes a centralized kill switch for big AI. Good luck with that. - Windows Central

AI is definitely going to kill us all, but letting a third party disable their GPUs remotely is probably not on the agenda for Space KKKaren and Winnie the Pooh. Anyway TLAs have probably already looked at it. Ya think Huaweis have backdoors and kill switches already? You betcha.